Hugging Face's logo

qualcomm
/
Qwen3-VL-4B-Instruct

Text Generation
PyTorch
llm
vlm
generative_ai
android
Model card Files
xet
 Community

Qwen3-VL-4B-Instruct: Optimized for Qualcomm Devices

Qwen3-VL is a vision-language model from Alibaba Cloud capable of understanding both text and images for multimodal reasoning tasks such as visual question answering and image captioning.

This is based on the implementation of Qwen3-VL-4B-Instruct found here. This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the Qualcomm® AI Hub Models library to export with custom configurations. More details on model performance across various devices, can be found here.

Qualcomm AI Hub Models uses Qualcomm AI Hub Workbench to compile, profile, and evaluate this model. Sign up to run these models on a hosted Qualcomm® device.

Deploying Qwen3-VL-4B-Instruct on-device

Follow the GenieX quickstart to install GenieX and deploy the model on a target device.

See the LLM-on-Genie tutorial to run with the Genie runtime. Note: Genie support will be deprecated soon.

Getting Started

There are two ways to deploy this model on your device:

Option 1: Download Pre-Exported Models

Below are pre-exported model assets ready for deployment.

Runtime Precision Chipset SDK Versions Download
GENIE w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile QAIRT 2.45 Download
GENIE w4a16 Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.45 Download
GENIE w4a16 Snapdragon® X2 Elite QAIRT 2.45 Download
GENIE w4a16 Snapdragon® X Elite QAIRT 2.45 Download
GENIE w4a16 Qualcomm® Dragonwing™ IQ-8275 QAIRT 2.45 Download
GENIE w4a16 Qualcomm® SA8775P QAIRT 2.45 Download
GENIE w4a16 Qualcomm® Dragonwing™ IQ-9075 QAIRT 2.45 Download
GENIEX_LLAMACPP q4_0 Universal Download
GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Snapdragon® X2 Elite QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Snapdragon® X Elite QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-8275 QAIRT 2.45 Download
GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-9075 QAIRT 2.45 Download

For more device-specific assets and performance metrics, visit Qwen3-VL-4B-Instruct on Qualcomm® AI Hub.

Option 2: Export with Custom Configurations

Use the Qualcomm® AI Hub Models Python library to compile and export the model with your own:

  • Custom weights (e.g., fine-tuned checkpoints)
  • Custom input shapes
  • Target device and runtime configurations

This option is ideal if you need to customize the model beyond the default configuration provided here.

See our repository for Qwen3-VL-4B-Instruct on GitHub for usage instructions.

Model Details

Model Type: Model_use_case.text_generation

Model Stats:

  • Model architecture: Transformer with ViT Vision Encoder, Grouped Query Attention (GQA), and SwiGLU activation.
  • Response Rate: Rate of response generation after the first response token.
  • Supported languages: 100+ languages and dialects
  • TTFT: Time To First Token is the time it takes to generate the first response token. This is expressed as a range because it varies based on the length of the prompt.

Performance Summary

Model Runtime Precision Chipset Context Length Response Rate (tokens per second) Time To First Token (range, seconds)
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 21.060619 0.9764346666666667 - 3.9057386666666667
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 20.998302 0.5379623333333333 - 2.1518493333333333
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 19.885145 0.11516533333333333 - 0.4606613333333333
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 5.96164 2.9840616774193545 - 95.48997367741934
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 13.994156 1.0128197741935483 - 32.410232774193545
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 10.74046 0.14600483870967743 - 4.672154838709678
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 512 21.01329 1.1112463333333333 - 4.444985333333333
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 512 19.826625 0.5111666666666667 - 2.0446666666666666
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 512 17.180041 0.12326733333333334 - 0.49306933333333336
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 4096 5.061711 2.960835548387097 - 94.7467375483871
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 4096 12.541984 1.1360922258064516 - 36.35495122580645
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® 8 Elite For Galaxy Mobile 4096 7.934172 0.19025909677419356 - 6.088291096774194
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 31.89261 1.117335 - 4.46934
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 27.412715 0.976281 - 3.905124
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 512 25.896353 0.399075 - 1.5963
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 18.488688 18.655281 - 596.968992
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 22.088114 15.776444999999999 - 504.84623999999997
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X2 Elite 4096 14.693486 4.205133 - 134.564256
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 21.596461 2.323608 - 9.294432
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 20.326375 2.2151439999999996 - 8.860575999999998
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 512 12.642968 1.004591 - 4.018364
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 5.788148 49.641781 - 1588.536992
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 11.385522 45.134884 - 1444.316288
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Snapdragon® X Elite 4096 6.078229 11.663933 - 373.245856
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 3.968657 15.613197 - 62.452788
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 2.530889 23.050572 - 92.202288
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 512 9.103716 1.280274 - 5.121096
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 1.292033 182.580826 - 5842.586432
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 1.932154 284.515082 - 9104.482624
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-8275 4096 4.635059 13.209513999999999 - 422.70444799999996
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 512 17.084362 6.884202 - 27.536808
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 512 7.898823 7.846550000000001 - 31.386200000000002
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 512 11.359227 1.175309 - 4.701236
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 21.596461 2.323608 - 9.294432
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 20.326375 2.2151439999999996 - 8.860575999999998
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 512 12.642968 1.004591 - 4.018364
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 21.01329 1.1112463333333333 - 4.444985333333333
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 19.826625 0.5111666666666667 - 2.0446666666666666
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 512 17.180041 0.12326733333333334 - 0.49306933333333336
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 4096 9.021444 88.038753 - 2817.240096
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 4096 5.696099 100.822643 - 3226.324576
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-9075 4096 5.250687 12.489227000000001 - 399.65526400000005
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 5.788148 49.641781 - 1588.536992
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 11.385522 45.134884 - 1444.316288
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ IQ-X7181 4096 6.078229 11.663933 - 373.245856
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 5.061711 2.960835548387097 - 94.7467375483871
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 12.541984 1.1360922258064516 - 36.35495122580645
Qwen3-VL-4B-Instruct GENIEX_LLAMACPP q4_0 Qualcomm® Dragonwing™ Q-8750 4096 7.934172 0.19025909677419356 - 6.088291096774194
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 512 15.998796 0.095944 - 0.383776
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 1024 15.560634 0.09915399999999999 - 0.7932319999999999
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 2048 15.505643 0.095107 - 1.521712
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile 4096 15.132583 0.103378 - 3.308096
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 512 15.937471 0.085935 - 0.34374
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 1024 15.460394 0.09641 - 0.77128
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 2048 15.467407 0.09730899999999999 - 1.5569439999999999
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile 4096 15.654413 0.098193 - 3.142176
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® X2 Elite 4096 39.34948 0.049329 - 1.578528
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Snapdragon® X Elite 4096 20.816945 0.099474 - 3.183168
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-8275 4096 16.014892 0.102327 - 3.274464
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 512 15.937471 0.085935 - 0.34374
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 1024 15.460394 0.09641 - 0.77128
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 2048 15.467407 0.09730899999999999 - 1.5569439999999999
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-9075 4096 18.063043 0.09700700000000001 - 3.1042240000000003
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-X7181 4096 20.816945 0.099474 - 3.183168
Qwen3-VL-4B-Instruct GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750 4096 15.654413 0.098193 - 3.142176
Vision-Encoder GENIE w4a16 Snapdragon® X Elite
Vision-Encoder GENIE w4a16 Qualcomm® Dragonwing™ IQ-X7181
Vision-Encoder GENIEX_QAIRT w4a16 Snapdragon® 8 Elite Gen 5 For Galaxy Mobile
Vision-Encoder GENIEX_QAIRT w4a16 Snapdragon® 8 Elite For Galaxy Mobile
Vision-Encoder GENIEX_QAIRT w4a16 Snapdragon® X Elite
Vision-Encoder GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ IQ-X7181
Vision-Encoder GENIEX_QAIRT w4a16 Qualcomm® Dragonwing™ Q-8750

License

  • The license for the original implementation of Qwen3-VL-4B-Instruct can be found here.

References

Community

Usage and Limitations

This model may not be used for or in connection with any of the following applications:

  • Accessing essential private and public services and benefits;
  • Administration of justice and democratic processes;
  • Assessing or recognizing the emotional state of a person;
  • Biometric and biometrics-based systems, including categorization of persons based on sensitive characteristics;
  • Education and vocational training;
  • Employment and workers management;
  • Exploitation of the vulnerabilities of persons resulting in harmful behavior;
  • General purpose social scoring;
  • Law enforcement;
  • Management and operation of critical infrastructure;
  • Migration, asylum and border control management;
  • Predictive policing;
  • Real-time remote biometric identification in public spaces;
  • Recommender systems of social media platforms;
  • Scraping of facial images (from the internet or otherwise); and/or
  • Subliminal manipulation
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
Text Generation
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Paper for qualcomm/Qwen3-VL-4B-Instruct